PRING Raw Data

This directory contains the raw data files used in the PRING benchmark.
These files can be used to regenerate the processed datasets or to extend the benchmark with additional species. Data processing scripts are available in the PRING GitHub repository.

---

1. Data Format

This directory includes two files:

• ppi.txt
  Contains protein–protein interaction (PPI) data. The format is:

uniprot_id_1    PPI    uniprot_id_2    data_source
P15112          PPI    Q55GJ7          string

• idmapping.fasta
  Contains protein sequences in FASTA format. Each entry begins with a header line (>), followed by the UniProt ID, metadata, and the amino acid sequence. Example:

> sp|P50399|GDIB_RAT Rab GDP dissociation inhibitor beta OS=Rattus norvegicus OX=10116 GN=Gdi2 PE=1 SV=2
> MNEEYDVIVLGTGLTECILSGIMSVNGKKVLHMDQNPYYGGESASITPLEDLYKRFKLPG
> ...
> YKRMTGSEFDFEEMKRKKNDIYGED

The OX field in the header specifies the NCBI taxonomy ID, which can be used to filter sequences by species.

---

2. Data Sources

The interactions in ppi.txt are aggregated from multiple curated databases, including:
STRING, IntAct, Reactome, and UniProt.

---

Citation

If you find PRING useful, please consider citing:

@article{zheng2025pring,
title={PRING: Rethinking Protein-Protein Interaction Prediction from Pairs to Graphs},
author={Zheng, Xinzhe and Du, Hao and Xu, Fanding and Li, Jinzhe and Liu, Zhiyuan and Wang, Wenkang and Chen, Tao and Ouyang, Wanli and Li, Stan Z and Lu, Yan and others},
journal={arXiv preprint arXiv:2507.05101},
year={2025}
}

@inproceedings{zheng2025pring,
title={{PRING}: Rethinking Protein-Protein Interaction Prediction from Pairs to Graphs},
author={Xinzhe Zheng and Hao Du and Fanding Xu and Jinzhe Li and Zhiyuan Liu and Wenkang Wang and Tao Chen and Wanli Ouyang and Stan Z. Li and Yan Lu and Nanqing Dong and Yang Zhang},
booktitle={The Thirty-ninth Annual Conference on Neural Information Processing Systems (NeurIPS) Datasets and Benchmarks Track},
year={2025},
url={https://openreview.net/forum?id=mHCOVlFXTw}
}

---

Further Information

For dataset regeneration instructions and additional details, please refer to the PRING GitHub repository.